I do, and a few times had to get my mechanical engineer to simplify 3D models, so that the KiCad STEP export didn’t cause Solidworks import to crash. I am sure Solidworks had a STEP import size bug.
FEM is a generic method for dividing some object (or 3D space) into very small sections, putting the thing in a (big) matrix and then running a lot of math to calculate interactions with the cells in the matrix. It is “just” a math thing, and it can be used for very varying applications. A few examples:
- Mechanical stress and material deformation. (Both elastic and plastic).
- Thermals (How heat moves though a PCB / Heatsink.
- Modelling of Electric and Magnetic fields
- Flame front propagation inside rocket motors.
Solidworks is the light and cheap entry level version of Dassaults offering. So it is on purpose kneecapped so they can sell Catia. So it has artificial limits that exhibit now and then.
Solidworks does indeed have a limitation on how many parts a document can have. Which means that the new export option to export all copper and vias as individual parts is going to be total killer for Solidworks. But this is indeed a artifical limitation of the software, not that solidworks as a software could not handle scenes this big (because it could before being aquired by dassault).
so does NX… one of the only true open standards for exchanging mechanical solid files is being neglected to push their own file formats (eg x_t). The last Kicad 4-card kicad design that I created STEP’s for to assist in mechanical packaging ended up taking about 1h per file to load… then convert to open “instantly”.
I decided to use step-up to put track information in not just to push how bad NX importer of a STANDARD was, but I needed to perform box-level creepage-clearance from screwheats to 1000V nodes…
1 Like
I’d like to add:
The B+ model (for D I don’t care) should have 0,0,0 at footprint center and orientation such to be able to use it directly (without any rotations needed). I hope for other PCB design software it means the same orientation.
1 Like
Yes, I agree producing borg models should be quite simple with the current scripting methods we are creating.
Regarding treating KiCAD seriously, we are already working on having all Würth products in KiCAD officially from us, as we think it makes a lot of sense.
A colleage of mine is working of extracting the footprints from our current models and pass them through the system and into the KiCAD repository.
11 Likes
Last summer I asked my local sales representative (in DK) for kicad footprints and he seemed genuinely interested in fulfilling that request.
Now reading this thread = 
I agree with B+ and D (voted D)
1 Like
your problems is really that your using freeCAD
I’d say the problem is moreso that he doesn’t know how to do that in freecad - which is simple as sketching a trace over the face and extruding “up to face” and clicking on the other side of it.
Personally, I don’t like cycloidal models in my stock models if I can help it because while a couple aren’t going to bog down a file, they’re always going to be extremely demanding geometry to render. Usually not a problem on my own workstation, but the odd time you need to work with a model on someone else’s (showing them some feature or whatever). Almost every time this happens to me, turning the fasteners off does the trick.
I can’t really think of a good reason for needing the detailed winding to be included since it’ll be canned anyway rather than the exact winding of whatever inductor. The same functional level of detail can be accomplished with a texture.
TL;DR: It’s easy to forget that not everyone has a thousand-core nitrous-oxide-fueled graphics workstation - until you have to use a dog-slow one for working on something important.
@ AlfVII
There’s also an environmental angle. The detailed windings by and large don’t add a value proposition - rendering them does consume substantial unnecessary electricity if you consider scope and scale. And we all know know that the nation which is going to be doing a lot of this rendering still burns a lot of coal.
People are expressly stating D because of the terminations. I think I can generalize the responses a little more clearly. End users are interested in the details that include features that will interact with what they are modelling.
As I mentioned previously, detailed windings are just eye candy. I’m sure some marketting folk would argue otherwise but to me they’re just cruft. I’d like to see not only the terminations but I would like to see keep out zones if any exist - creepage and clearance indicators would be usefu,l for example. Some indication of any conductive bondable surfaces would also be useful. Another detail I haven’t seen mentioned would be indicated EMI radiation planes. Sensitivity zones for hall sensors and current transformers could also be very useful. One final suggestion would be to include surfaces that map lines of force extending from poles: pole pieces, gapped cores and shaped cores - when those take part in a design, they are often vital.
Those are details I sometimes add to my models and they may already exist in yours internally. Those details add a lot of value to the 3D models, IMHO.
2 Likes
I found significant slowdowns in freecad when I had a lot of holes modeled with proper threads. An option is a fake thread, which renders way faster than a helical thread. Perhaps a good option to model coils:
For an inductor model, I definitely want the smt pins – that is a must-have. A cylinder is fine for the coil (and can be copper-colored), but simplifed non-helical coils would also be nice.
1 Like
The reason you would need it if you really want to simulate in some muliphysics simulator which sometimes comes up for me. Its annoying to have to model most parts again if you need this.
The reason for wanting choice “D” has been for FEM analysis. This has been clearly stated several times in this thread. I would guess that you are unfamiliar with this based on the content of your text.
It is ironic that you later ask for things like radiation planes and field lines, since these are exactly the things you would calculate from model “D”. Furthermore, these latter things are in fact not clear cut, and it is in fact the interaction of multiple components that is normally determined, hence it is not really that useful (or possible) to accurately calculate them for standalone components, especially magnetic components. It is done on the whole PCB or a significant section of the PCB.
Several people have presented an appropriate solution:
B+termination for PCB CAD
D for 3D FEM simulations
In fact, the vendor could make available both models relatively easily, and the end user can choose. Most will choose B+, because it’s only a small (but growing) subset of people using FEM analysis. But, if the vendor is making D available, it is a huge benefit to those who need to do FEM.
John
1 Like
Definitely D. Computers are always getting faster and there is a huge push on improving render speeds and GPU performance, Storage is not an issue. More details in the model generates more accurate results.
It seems I should have allowed the poll to run for much longer. There is still a lot of interest in this thread.
Unfortunately polls are only able to be changed for the first five minutes after creation, which is a bit of a bummer. 
Computer speed ups have become far smaller these days and only obvious if the task is threaded across several cpus.
Storage is still a bit tight on a 250 GB laptop and a big file becomes a nuisance for sharing via email.
Most 3D models are used for mechanical assembly work.
Do you have any idea how big a detailed spiral wound transformer STEP model can be?
I just checked, model D is 700 kb, but it is a simple magnetic. I am just guessing, but maybe a more complex one, with several windings, can be up to a few Mb.
Model will definitely be more complicated for sure. Poorly designed models can be huge. The OP did not specify the difference so it leads us to make assumptions and that is unfortunate. It is advisable that they construct the model with care and in an efficient manner so loading and rendering are minimized.
@steves, the OP is doing everything you see in the pictures by script, with the number of segments of everything a parameter to configure 
My idea is to automate everything so it can be configured and generated on the fly when an user request the model.
Just for throwing a big bag of grains in the chicken-shack… 
Example of how other manufacturers do it
ESP32C3-WROOM “official” 3D
And this after removing the screening cap, quite complex isn’t it?
from What is the difference between step and wrl 3d models?
TlDr: use wrl to make pictures for your marketing guy, use step to exchange information with your fellow mechanical engineers.
And wrl files are much smaller!
But I did not test .wrl in Kicad yet as wrl does not seem to render in 8.0.1… maybe bug?
I prefer to work with models that are as realistic as possible. You could provide space filler models as a bonus for those that want them.
More importantly when will you create kicad libraries of your parts?
Even more importantly, all models should be put on the XY plane centred around 0,0. I have no patience for people whole supply models on a Z plane and totally off centre, that is several component sizes away from the centre, are your CAD modellers insane?. I always model around the centre of the plane. It is much easier that way. The parts usually have symetry, if you don’t use the planes you just have to add your own centre axis to the model, why do you insist on hurting yourselves like this?
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.